Impelementasi Chatbot Kesehatan Mental berbasis LLM dengan Integrasi Emotion Recognition dan Tes MBTI untuk Optimalisasi Layanan Psikologi

Arismana, Fina Keiza (2025) Impelementasi Chatbot Kesehatan Mental berbasis LLM dengan Integrasi Emotion Recognition dan Tes MBTI untuk Optimalisasi Layanan Psikologi. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5027211028-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (7MB) | Request a copy

Abstract

Masalah kesehatan mental secara global menjadi perhatian utama, dengan laporan World Health Organization (WHO) yang menunjukkan peningkatan signifikan kasus gangguan mental, terutama di kalangan mahasiswa. Stigma, keterbatasan biaya, dan kurangnya informasi menjadi hambatan dalam mengakses layanan profesional. Penelitian ini mengusulkan pengembangan chatbot kesehatan mental berbasis Large Language Model (LLM) yang mampu memberikan respons empatik dengan mempertimbangkan emosi dan tipe kepribadian pengguna. Sistem ini mengintegrasikan LLaMA 3 untuk percakapan, RoBERTa untuk deteksi emosi, dan tes MBTI berbasis rule-based dengan pendekatan Forward Chaining. Chatbot dirancang berbasis sesi, menyimpan riwayat percakapan dan otomatis menghasilkan ringkasan yang dapat dimanfaatkan dalam sesi konseling profesional. Pengujian mencakup evaluasi performa model, uji fungsionalitas fitur, dan user testing terhadap mahasiswa dan psikolog. Hasil menunjukkan sistem mampu menghasilkan respons kontekstual dengan akurasi deteksi emosi mencapai 97% dan akurasi rata-rata BERTScore-F1 pada model chatbot mencapai 87%. Selain itu, hasil dari 28 responden dengan latar belakang umur, angkatan, dan jurusan yang berbeda, chatbot dinilai mudah digunakan dan dapat diterima dengan baik, meskipun beberapa respons masih perlu disempurnakan. Penelitian ini menunjukkan bahwa integrasi LLM, deteksi emosi, dan MBTI efektif dalam membangun chatbot yang dapat mendukung optimalisasi layanan psikologis, khususnya sebagai media awal dalam membantu penanganan isu kesehatan mental.
=================================================================================================================================
Mental health issues have become a major global concern, with reports from the World Health Organization (WHO) indicating a significant rise in mental disorder cases, particularly among university students. Stigma, financial limitations, and lack of information are key barriers to accessing professional mental health services. This research proposes the development of a mental health chatbot based on a Large Language Model (LLM) capable of providing empathetic responses by considering users’ emotions and personality types. The system integrates LLaMA 3 for conversational capabilities, RoBERTa for emotion detection, and a rule-based MBTI test using the Forward Chaining approach. The chatbot is session-based, capable of saving conversation histories and automatically generating conversation summaries to support professional counseling sessions. The evaluation covers model performance testing, feature functionality validation, and user testing involving students and psychologists. Results indicate that the system can generate contextually appropriate responses, with an emotion detection accuracy of 97% and an average accuracy BERTScore-F1 for model chatbot of 87%. Additionally, based on responses from 27 participants with varying ages, academic years, and study programs, the chatbot was considered user-friendly and well-accepted, despite the need for improvement in certain responses. This research demonstrates that the integration of LLM, emotion detection, and MBTI is effective in developing a chatbot that supports the optimization of psychological services, particularly as an initial medium to assist in addressing mental health issues.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Chatbot, Kesehatan Mental, LLaMA, MBTI, RoBERTa, Mental Health
Subjects:	T Technology > T Technology (General) T Technology > T Technology (General) > T58.8 Productivity. Efficiency T Technology > T Technology (General) > T59.7 Human-machine systems.
Divisions:	Faculty of Information and Communication Technology > Information Technology > 59201-(S1) Undergraduate Thesis
Depositing User:	Fina Keiza Arismana
Date Deposited:	10 Jul 2025 03:37
Last Modified:	10 Jul 2025 03:37
URI:	http://repository.its.ac.id/id/eprint/119465

Actions (login required)

View Item